WeMo MCP Server

<!-- Use this file to provide workspace-specific custom instructions to Copilot. For more details, visit https://code.visualstudio.com/docs/copilot/copilot-customization#_use-a-githubcopilotinstructionsmd-file --> # WeMo MCP Server - Copilot Instructions ## 🤖 Mandatory Rules for AI Agents > These rules apply to GitHub Copilot and any AI assistant making changes in this repo. > Violating them has caused CI failures and broken releases in the past. ### BEFORE EVERY `git commit` — no exceptions 1. **Run the full test suite** and confirm it passes: ```bash .venv/bin/python -m pytest tests/test_server.py tests/test_phase2.py tests/test_models.py -q --tb=short # Must show: X passed, 0 failed ``` - This takes ~5 seconds. There is no acceptable reason to skip it. - Do NOT use `--no-verify`. Pre-commit hooks run black, ruff, mypy, and pytest automatically. - If a test fails, fix the test or the code before committing. Do not commit broken tests. 2. Only after the test suite passes, commit normally (no `--no-verify`): ```bash git commit -m "your message" ``` The pre-commit hooks will run black, ruff, mypy, and pytest automatically. If any hook fails, fix the issue and retry. ### BEFORE EVERY `git push` — always ask the user first Show the commit summary and wait for explicit confirmation before pushing: ```bash git log origin/main..HEAD --oneline ``` Exception: user has explicitly said "push" or "commit and push" in their request. ### WHY THIS MATTERS The `cache_keys` removal (Feb 21 2026) broke CI because tests were not run before committing. A 5-second local test run would have caught it immediately. **Never let a CI failure be the first signal that something is broken.** **Production-ready MCP Server for WeMo smart home device control via AI assistants.** ## Project Identity - **Repository**: https://github.com/apiarya/wemo-mcp-server - **Package**: `wemo-mcp-server` (PyPI) - **MCP Name**: `io.github.apiarya/wemo` - **Version**: v1.4.0 (stable) - **Status**: Production • Published on PyPI and MCP Registry ## Project Type & Architecture - **Type**: Model Context Protocol (MCP) Server - **Framework**: FastMCP (mcp>=1.2.0) - **Language**: Python 3.10+ with type hints - **Transport**: stdio (JSON-RPC over stdin/stdout) - **Package Manager**: uv (with uv.lock for reproducible builds) - **Build System**: Hatchling ## Core Functionality Natural language control of WeMo smart home devices through AI assistants: - Multi-phase device discovery (UPnP/SSDP + network scanning) - Real-time device control (on/off/toggle/brightness) - Device management (rename, HomeKit codes) - Status monitoring - Automatic device caching ## Project Structure ``` wemo-mcp-server/ ├── src/wemo_mcp_server/ │ ├── __init__.py # Package init and version (__version__ = "1.4.0") │ ├── __main__.py # Entry point for python -m wemo_mcp_server │ ├── server.py # Main server (~1570 lines) - tools, resources, prompts, elicitations │ ├── cache.py # Persistent JSON device cache (DeviceCache) │ ├── config.py # YAML + env var configuration management │ ├── error_handling.py # Retry decorator + error classification │ └── models.py # Pydantic request/response models ├── tests/ │ ├── __init__.py │ ├── test_server.py # Unit tests (30 tests, server.py) │ ├── test_phase2.py # Unit tests (72 tests, cache/config/error_handling) │ ├── test_models.py # Unit tests (Pydantic models) │ └── test_e2e.py # End-to-end tests with real devices ├── .github/ │ ├── workflows/ │ │ ├── ci.yml # CI: multi-version tests + quality checks │ │ └── release.yml # Release: publish to PyPI + MCP Registry │ ├── copilot-instructions.md # This file │ ├── PYPI_PUBLISHING.md # PyPI trusted publishing docs │ └── PUBLISHING_AUTOMATION.md # Automation guide ├── assets/ │ └── claude-example.png # Screenshot for README ├── pyproject.toml # Package config (version, dependencies, metadata) ├── server.json # MCP Registry metadata ├── uv.lock # Locked dependencies (291KB) ├── README.md # Comprehensive docs (456 lines) ├── CHANGELOG.md # Version history ├── LICENSE # MIT License ├── MIGRATION.md # Migration documentation ├── RELEASE.md # Release process guide ├── RELEASE_CHECKLIST.md # Quick release checklist ├── RELEASE_NOTES_v1.1.0.md # v1.1.0 release notes ├── RELEASE_NOTES_v1.1.1.md # v1.1.1 release notes └── MCP_REGISTRY_SUBMISSION.md # Registry submission docs ``` ## Recent Improvements - Phase 1 Complete ✅ **Quality Foundation (Weeks 1-5)** - Completed February 21, 2026 ### CI/CD Pipeline - ✅ GitHub Actions workflow with multi-version testing (Python 3.10-3.13) - ✅ Automated quality checks (pytest, ruff, black, isort) - ✅ Pre-commit hooks configured (`.pre-commit-config.yaml`) - ✅ Coverage reporting integrated - ✅ All checks passing on every push ### Test Suite Expansion - **Before**: 12 tests, 32.14% coverage - **After**: 30 tests, 78.14% coverage - **Improvement**: +18 tests, +46pp coverage - **Test Speed**: ~3.5 seconds for full suite - **New Test File**: 580 lines with comprehensive coverage ### Code Quality Improvements - ✅ Reduced code complexity: - `scan_subnet()`: Complexity 13 → 6 (extracted 4 helper methods) - `control_device()`: Complexity 13 → 6 (extracted 4 helper functions) - ✅ All complexity violations resolved (C901, PLR0912) - ✅ All docstring checks passing (D107, D205, D401) - ✅ 22+ pragmatic linting rules configured - ✅ Code formatted with black, imports sorted with isort ### Helper Functions Extracted **WeMoScanner methods:** - `_run_upnp_discovery()` - Phase 1 UPnP/SSDP discovery - `_parse_cidr_network()` - CIDR validation and parsing - `_probe_active_ips()` - Phase 2 parallel port probing - `_verify_wemo_devices()` - Phase 3 device verification **Module-level helpers:** - `_validate_action()` - Action parameter validation - `_validate_brightness()` - Brightness range validation - `_perform_device_action()` - Execute device actions - `_build_control_result()` - Build result dictionaries ### Local Validation Workflow - ✅ Comprehensive pre-push checklist - ✅ Quick validation script (`scripts/validate.sh`) - ✅ Shift-left testing philosophy enforced - ✅ All checks must pass locally before push ### Benefits Achieved - 🚀 **Faster Development**: Instant local feedback, no CI wait time - 🛡️ **Higher Quality**: 78% test coverage, automated enforcement - 📈 **Better Maintainability**: Reduced complexity, clear structure - ✅ **Professional Standards**: CI/CD pipeline, comprehensive testing ## Recent Improvements - Phase 3 Complete ✅ **MCP Primitives (v1.4.0)** - Completed February 22, 2026 ### New MCP Primitives - ✅ MCP Resources: `devices://` (static index) + `device://{device_id}` (live state, URL-decoded) - ✅ MCP Prompts: 4 guided prompts (discover-devices, device-status-report, activate-scene, troubleshoot-device) - ✅ MCP Elicitations: subnet confirmation in `scan_network` + device disambiguation in `control_device` ### Infrastructure - ✅ CI/CD: multi-version (3.10–3.13), Dependabot weekly scans, pip-audit security checks - ✅ Branch protection: 6 required status checks, no force-push/delete, auto-merge enabled - ✅ Release gate: `release.yml` blocks publish if any CI check failed on the tagged commit - ✅ 128 tests across 3 test files, ~82% overall coverage - ✅ Release v1.4.0 published to PyPI + MCP Registry ## MCP Tools (9 Total) - server.py All tools are async and decorated with `@mcp.tool()`: ### 1. `scan_network(subnet, timeout, max_workers)` - **Purpose**: Discover WeMo devices on network - **Method**: Multi-phase (UPnP/SSDP first, then port scanning backup) - **Returns**: Device list with full details + caches devices - **Performance**: ~23-30s for full /24 subnet with 60 workers ### 2. `list_devices()` - **Purpose**: List cached devices from previous scans - **Returns**: Device count and device list - **Note**: Falls back to persistent JSON cache on restart — no rescan required ### 3. `get_device_status(device_identifier)` - **Purpose**: Get real-time device state - **Input**: Device name or IP address - **Returns**: State (on/off), brightness (if dimmer), device info - **Note**: Auto-reconnects from file cache if in-memory cache is empty ### 4. `control_device(device_identifier, action, brightness)` - **Purpose**: Control devices - **Actions**: "on", "off", "toggle", "brightness" - **Brightness**: 1-100 for dimmers - **Returns**: Success status, new state, brightness - **Note**: Auto-reconnects from file cache if in-memory cache is empty ### 5. `rename_device(device_identifier, new_name)` - **Purpose**: Change device friendly name - **Updates**: Device cache and WeMo device itself - **Returns**: Old name, new name, success status ### 6. `get_homekit_code(device_identifier)` - **Purpose**: Extract HomeKit setup code - **Note**: Not all devices support HomeKit - **Returns**: HomeKit code (format: XXX-XX-XXX) ### 7. `get_cache_info()` - **Purpose**: Inspect persistent cache status - **Returns**: Cache age, TTL, device count, expiry status ### 8. `clear_cache()` - **Purpose**: Clear file + in-memory cache to force fresh scan - **Returns**: Success status ### 9. `get_configuration()` - **Purpose**: View all current config settings - **Returns**: Network, cache, and logging settings with active values ## MCP Resources (2 Total) - server.py Decorated with `@mcp.resource()`: ### `devices://` - **Purpose**: Static JSON index of all devices in the in-memory cache - **Returns**: List of `{name, ip, type}` for every cached device ### `device://{device_id}` - **Purpose**: Live state for a specific device (name or IP, URL-decoded) - **Returns**: Current state (on/off), brightness (if dimmer), model info ## MCP Prompts (4 Total) - server.py Decorated with `@mcp.prompt()`: | Prompt | Description | |--------|-------------| | `discover-devices` | Guided network scan with subnet selection | | `device-status-report` | Full status summary of all cached devices | | `activate-scene` | Control multiple devices simultaneously as a scene | | `troubleshoot-device` | Step-by-step diagnostic for a non-responsive device | ## MCP Elicitations Triggered automatically when context is ambiguous. Schemas defined in `server.py` lines 43–55: | Schema | Used by | Triggers when | |--------|---------|---------------| | `_SubnetChoiceSchema(subnet: str)` | `scan_network` | Default `192.168.1.0/24` with no `WEMO_MCP_DEFAULT_SUBNET` env override | | `_DeviceChoiceSchema(device_name: str)` | `control_device` | Device identifier not found in in-memory or file cache | **Client support**: Claude Desktop v1.1+ ✅ · MCP Inspector v0.20+ ✅ · VS Code ❌ ## Development Guidelines ### Code Style - **Async/await**: All MCP tools must be async - **Type hints**: Use throughout (strict mypy) - **Logging**: Use `logging` module, log to stderr only (never stdout) - **Error handling**: Try/except with informative error dictionaries - **Docstrings**: Google-style docstrings for all functions ### MCP Server Patterns - Use `FastMCP` for server initialization: `mcp = FastMCP("wemo-mcp-server")` - Tools return Dict[str, Any] with consistent structure - Include "error" key in results when operations fail - Provide "suggestion" keys for common issues - Cache devices in global `_device_cache` dict ### Network Operations - Use `pywemo` library for WeMo device interaction - Run sync operations in thread pool: `loop.run_in_executor(None, func)` - Keep network scanning non-invasive (timeout 0.6s default) - Parallel scanning with ThreadPoolExecutor (60 workers default) - Multi-phase discovery: UPnP/SSDP primary, port scanning backup ### Version Management **CRITICAL**: Keep versions synchronized across 3 files: 1. `pyproject.toml` - Line 7: `version = "X.Y.Z"` 2. `src/wemo_mcp_server/__init__.py` - Line 3: `__version__ = "X.Y.Z"` 3. `server.json` - Line 10: `"version": "X.Y.Z"` GitHub Actions workflow verifies version consistency before publishing. ## Testing ### Unit Tests - **128 tests** across 3 test files (30 + 72 + 26) - **~81.88% overall coverage** - All tests pass in ~4 seconds - Run: `pytest tests/test_server.py tests/test_phase2.py tests/test_models.py -v` ### Test Coverage Breakdown ``` src/wemo_mcp_server/__init__.py 100.00% coverage src/wemo_mcp_server/error_handling.py 100.00% coverage src/wemo_mcp_server/cache.py 96.34% coverage src/wemo_mcp_server/config.py 91.11% coverage src/wemo_mcp_server/models.py 86.05% coverage src/wemo_mcp_server/server.py ~57% coverage ────────────────────────────────────────────────────── TOTAL ~81.88% coverage ``` ### E2E Tests (test_e2e.py) - Requires actual WeMo devices on network - Tests all 9 MCP tools end-to-end - Configurable device count and control testing - **Not run in CI** (requires physical hardware) - Run: `python tests/test_e2e.py` ### Test Commands ```bash # Run all unit tests (fast, CI-compatible) — REQUIRED before every commit .venv/bin/python -m pytest tests/test_server.py tests/test_phase2.py tests/test_models.py -q --tb=short # Run with coverage pytest tests/test_server.py tests/test_phase2.py tests/test_models.py --cov=wemo_mcp_server --cov-report=html # Run E2E tests (needs real devices) python tests/test_e2e.py # Quick validation (all checks) ./scripts/validate.sh ``` ## Common Development Commands ```bash # Run server locally uvx wemo-mcp-server # or python -m wemo_mcp_server # Run tests pytest tests/ -v # Format code black src/ tests/ isort src/ tests/ # Type check mypy src/ # Lint ruff check src/ tests/ # Build package python -m build # Install locally for testing pip install -e . ``` ## Commit Practices ### Pre-Commit Security Check 🔒 CRITICAL **Before every commit**, scan for sensitive data and secrets: #### What to Check: 1. **API Keys & Tokens** - [ ] No API keys (OpenAI, Anthropic, etc.) - [ ] No access tokens (GitHub, PyPI, etc.) - [ ] No authentication credentials 2. **Configuration Files** - [ ] No hardcoded passwords or secrets - [ ] No private keys or certificates - [ ] No database credentials 3. **Personal Information** - [ ] No email addresses (except public ones in documentation) - [ ] No phone numbers or addresses - [ ] No internal URLs or endpoints 4. **MCP Registry Tokens** - [ ] No `.mcpregistry_github_token` content - [ ] No `.mcpregistry_registry_token` content - [ ] These files should be in `.gitignore` 5. **Environment Files** - [ ] No `.env` files committed - [ ] No secrets in shell scripts - [ ] No hardcoded credentials in config files #### Quick Security Scan Commands: ```bash # Check for common secret patterns git diff --cached | grep -iE '(api[_-]?key|secret|token|password|credential)' # Check for potential secrets in all staged files git diff --cached --name-only | xargs grep -iE '(api[_-]?key|secret|token|password|credential)' 2>/dev/null # Check for email addresses git diff --cached | grep -E '[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}' # View all staged changes for manual review git diff --cached ``` #### Common False Positives (OK to commit): - ✅ Example API keys in documentation (clearly marked as examples) - ✅ Public PyPI tokens in documentation (for educational purposes) - ✅ "secret" or "token" in variable names or comments - ✅ Public email addresses in LICENSE or CONTRIBUTORS #### If You Find Secrets: ```bash # Unstage the file git reset HEAD <file> # Remove the secret from the file vim <file> # Stage the cleaned file git add <file> ``` **⚠️ If already committed but not pushed:** ```bash # Amend the last commit git commit --amend # Or reset and recommit git reset HEAD~1 # Clean files, then re-commit ``` **🚨 If already pushed:** 1. Immediately revoke the exposed credential 2. Generate a new credential 3. Update the code with new credential (via environment variables) 4. Consider using `git filter-branch` or BFG Repo-Cleaner to remove from history 5. Notify team/users if necessary ### Pre-Commit Documentation Check ⚠️ REQUIRED **Before every commit**, verify if documentation needs updates: #### Files to Check: 1. **README.md** - [ ] Are new features documented? - [ ] Are installation instructions still accurate? - [ ] Are example prompts up-to-date? - [ ] Are MCP tool descriptions current? 2. **CHANGELOG.md** - [ ] Is the change documented under appropriate version? - [ ] Are breaking changes clearly marked? - [ ] Is the change categorized correctly (Added/Changed/Fixed/etc.)? 3. **Copilot Instructions** (`.github/copilot-instructions.md`) - [ ] Does code structure match documented structure? - [ ] Are new tools/functions documented? - [ ] Are dependencies list current? - [ ] Are development commands accurate? 4. **Version Files** (if version bump) - [ ] `pyproject.toml` - version field - [ ] `src/wemo_mcp_server/__init__.py` - __version__ - [ ] `server.json` - version field - [ ] All three must match exactly! 5. **Tool Docstrings** (in `server.py`) - [ ] Do docstrings match actual behavior? - [ ] Are parameters documented correctly? - [ ] Are return values accurate? #### Quick Check Command: ```bash # Before committing, ask yourself: # "What documentation might be affected by my changes?" # Then review relevant files: git diff README.md git diff CHANGELOG.md git diff .github/copilot-instructions.md git diff server.json pyproject.toml src/wemo_mcp_server/__init__.py ``` #### Common Documentation Updates: - **New MCP Tool** → Update README.md (tools section), copilot-instructions.md - **Bug Fix** → Update CHANGELOG.md - **New Dependency** → Update README.md, pyproject.toml, copilot-instructions.md - **API Change** → Update README.md, tool docstrings, CHANGELOG.md (breaking changes) - **Version Bump** → Update 3 version files + CHANGELOG.md - **New Feature** → Update README.md, CHANGELOG.md, copilot-instructions.md #### Documentation Debt If you can't update documentation immediately: ```bash # Create a TODO comment in the commit message git commit -m "Add feature X TODO: Update README.md with new feature documentation TODO: Add example prompts for feature X" ``` Then create a GitHub issue to track the documentation update. ### Shift Left: Local Validation 🚀 CRITICAL **Run ALL checks locally BEFORE committing or pushing!** **Philosophy**: Catch issues early, fail fast locally, never waste time with CI failures. #### Complete Pre-Commit/Pre-Push Checklist **ALWAYS run these checks locally before committing:** ```bash # 1. Run all tests (should take ~3-5 seconds) pytest tests/test_server.py -v --tb=short # Expected: 30 passed in ~3.5s # 2. Run linting (should be instant) ruff check src/ tests/ # Expected: All checks passed! # 3. Check code formatting (should be instant) black --check src/ tests/ # Expected: All done! ✨ 🍰 ✨ (6 files would be left unchanged) # 4. Check import sorting (should be instant) isort --check-only src/ tests/ # Expected: (no output = success) # 5. Verify module imports (should be instant) python -c "import wemo_mcp_server.server; print('✅ Module imports successfully')" # Expected: ✅ Module imports successfully ``` #### Quick Validation Script **Create this as `scripts/validate.sh` for one-command validation:** ```bash #!/bin/bash # Quick local validation script set -e # Exit on first error echo "🧪 Running tests..." pytest tests/test_server.py -v --tb=short echo "" echo "🔍 Running linting..." ruff check src/ tests/ echo "" echo "🎨 Checking code formatting..." black --check src/ tests/ echo "" echo "📦 Checking import sorting..." isort --check-only src/ tests/ echo "" echo "✅ Verifying module imports..." python -c "import wemo_mcp_server.server; print('✅ Module imports successfully')" echo "" echo "✅✅✅ All local checks passed! Safe to commit/push. ✅✅✅" ``` **Usage:** ```bash # Make executable chmod +x scripts/validate.sh # Run before every commit/push ./scripts/validate.sh ``` #### Auto-fix Common Issues If checks fail, auto-fix formatting issues: ```bash # Auto-fix formatting black src/ tests/ # Auto-fix import sorting isort src/ tests/ # Auto-fix some linting issues ruff check src/ tests/ --fix # Re-run validation ./scripts/validate.sh ``` #### Integration with Git Workflow **Recommended workflow:** ```bash # 1. Make code changes vim src/wemo_mcp_server/server.py # 2. Run tests first (fast check before committing) .venv/bin/python -m pytest tests/test_server.py tests/test_phase2.py tests/test_models.py -q --tb=short # 3. Commit — pre-commit hooks run automatically (black, ruff, mypy, pytest) git add -A git commit -m "Your commit message" # Hooks auto-fix formatting; if they modify files, stage and commit again # 4. Push (after user confirmation) git push origin main ``` #### Why This Matters **Time saved from CI failures:** - Local check: ~5 seconds - CI failure + debug + fix + repush: ~5-10 minutes - **Savings: 60-120x faster feedback loop** **Benefits:** - ✅ Instant feedback (no waiting for CI) - ✅ Catch issues before they reach origin - ✅ Cleaner git history (fewer "fix CI" commits) - ✅ Less context switching - ✅ Respect for CI resources - ✅ Professional development practice **Golden Rule**: If it won't pass CI, don't push it. Run checks locally first! ### Before Pushing 🚨 ALWAYS ASK **Never push without user confirmation!** Before executing `git push`: 1. **Review what will be pushed**: ```bash # See commits that will be pushed git log origin/main..HEAD --oneline # See all changes that will be pushed git diff origin/main..HEAD ``` 2. **Final security check**: ```bash # Scan all commits being pushed for secrets git log origin/main..HEAD -p | grep -iE '(api[_-]?key|secret|token|password|credential)' ``` 3. **Ask the user**: - Show summary of commits being pushed - Show file changes summary - Wait for explicit confirmation before pushing - Example: "Ready to push 3 commits (modified: 5 files). Proceed? (y/n)" 4. **Only push after confirmation**: ```bash # User says yes git push origin main # User says no # Explain what needs to be fixed and wait for next instruction ``` **⚠️ IMPORTANT**: This is a safety mechanism to prevent accidental pushes of: - Unreviewed code - Sensitive data that passed initial checks - Incomplete work - Commits to wrong branch **Exception**: Only skip confirmation for: - Automated CI/CD workflows (GitHub Actions) - Explicitly requested immediate pushes ("push now", "commit and push") ## Publishing & CI/CD ### Automated Publishing (✅ ACTIVE) **GitHub Actions workflow**: `.github/workflows/release.yml` **Process**: Create GitHub release → Automated publish to PyPI + MCP Registry **Features**: - ✅ PyPI publishing via trusted publishing (OIDC) - ✅ MCP Registry publishing via github-oidc authentication - ✅ Version verification (pyproject.toml + server.json) - ✅ No secrets/tokens needed (all OIDC-based) - ✅ Detailed release summary **Security**: All publishing uses short-lived OIDC tokens (no API keys) ### Manual Release Process ```bash # 1. Update versions (3 files) vim pyproject.toml src/wemo_mcp_server/__init__.py server.json # 2. Update CHANGELOG.md vim CHANGELOG.md # 3. Commit and push git add -A git commit -m "Release vX.Y.Z" git push origin main # 4. Create tag and GitHub release git tag vX.Y.Z git push origin vX.Y.Z # 5. Create release on GitHub # → Workflow publishes to PyPI + MCP Registry automatically! ``` See `.github/PUBLISHING_AUTOMATION.md` for comprehensive automation guide. ## MCP Client Testing ### Claude Desktop ```json // ~/Library/Application Support/Claude/claude_desktop_config.json { "mcpServers": { "wemo": { "command": "uvx", "args": ["wemo-mcp-server"] } } } ``` ### VS Code ```json // ~/.vscode/mcp.json { "servers": { "wemo": { "type": "stdio", "command": "uvx", "args": ["wemo-mcp-server"] } } } ``` Test prompts: - "Scan for WeMo devices on my network" - "Turn on the office light" - "What's the brightness of the bedroom dimmer?" ## Key Dependencies ### Production - `mcp>=1.2.0` - Model Context Protocol SDK - `httpx>=0.25.0` - Async HTTP client - `pywemo>=1.4.0` - WeMo device control library ### Development - `pytest>=7.0.0` - Testing framework - `pytest-asyncio>=0.21.0` - Async test support - `black>=23.0.0` - Code formatter - `isort>=5.12.0` - Import sorter - `mypy>=1.0.0` - Type checker - `ruff>=0.1.0` - Fast linter ## Important Notes ### Device Cache - Global `_device_cache` dict stores discovered devices - Keyed by both device name and IP address - Persists during server lifetime - Cleared on server restart ### Logging - **CRITICAL**: Only log to stderr (not stdout) - stdout is reserved for MCP JSON-RPC protocol - Use `logger.info()`, `logger.error()`, etc. - Configured in server.py with stderr stream ### Error Handling - Always return Dict with "error" key on failures - Include "suggestion" for common issues - List "available_devices" when device not found - Use try/except blocks for all external operations ### Network Discovery - UPnP/SSDP is primary method (most reliable) - Port scanning (49152-49155) is backup - Parallel probing with 60 concurrent workers - Configurable timeout (default 0.6s per port) ## Documentation Standards - Keep README.md current with features and installation - Update CHANGELOG.md for every release - Document breaking changes prominently - Include example prompts in tool docstrings - Maintain comprehensive release notes ## Git Workflow - **Main branch**: Production-ready code - **Tags**: Version format `vX.Y.Z` (e.g., `v1.1.1`) - **Commits**: Descriptive messages with context - **Pre-commit checks**: Always verify documentation needs updates (see Commit Practices section) - **No direct commits to main**: Use feature branches for major changes ## Support & Community - **Issues**: https://github.com/apiarya/wemo-mcp-server/issues - **Discussions**: GitHub Discussions tab - **PyPI**: https://pypi.org/project/wemo-mcp-server/ - **MCP Registry**: https://registry.modelcontextprotocol.io/?q=apiarya/wemo ## Related Projects - **pywemo**: https://github.com/pywemo/pywemo (underlying library) - **WeMo Ops Center**: https://github.com/qrussell/wemo-ops-center (desktop app) --- **Last Updated**: February 22, 2026 (v1.4.0 + MCP Resources, Prompts, Elicitations + mandatory AI pre-commit rules)